VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models | Read Paper on Bytez